7 Ways to Use Point Clouds to Improve Deformation Survey Accuracy | Practical On-site Tips

In the maintenance and management of bridges, tunnels, slopes, retaining walls, pavements, building exteriors, cultural heritage, and various infrastructure, the importance of condition surveys is increasing year by year. It is required to detect abnormalities such as cracks, settlement, bulging, deflection, spalling, steps, and deformation without overlooking them, record them in a reproducible manner, and enable stakeholders to evaluate them using the same criteria. However, in actual fieldwork it is also true that accuracy tends to vary due to differences in inspectors’ experience, variation in measurement locations, inconsistent recording methods, and reduced visibility caused by site conditions.

What is drawing attention, then, is the use of point clouds. Because point clouds can record an object's surface shape as a large number of points in three dimensions, they make it easier to capture subtle shape changes that conventional photographs or manual measurements could not fully pick up. Another major strength is that once data have been acquired, they can be reviewed later and validated from different perspectives. In the field of deformation surveys, the goal is not simply to create 3D representations; what matters is how point clouds are used to reduce overlooked deformations, maintain comparable records, and increase the objectivity of assessments.

This article provides an organized explanation of seven practical points for using point clouds to improve the accuracy of deformation surveys, aimed at practitioners searching for "変状調査 点群". It delves not only into the thinking behind implementation but also into common on-site stumbling blocks and operational perspectives to prevent a loss of accuracy.

Table of Contents

• Why Point Clouds Are Attracting Attention in Deformation Surveys

• Point Cloud Method 1 to Improve Deformation Survey Accuracy: Standardize and Record the Overall Shape

• Point Cloud Applications to Improve Deformation Survey Accuracy — 2: Sharing Deformation Locations Using Coordinates

• Point cloud utilization method 3 to improve the accuracy of deformation surveys: Confirm deformation amounts in cross-sections

• Point cloud application method 4 to improve the accuracy of deformation surveys: determine progression by time-series comparison

• Point cloud utilization method 5 for improving deformation survey accuracy: enhance interpretation accuracy by combining with photographs

• Point Cloud Utilization Method 6 to Improve Deformation Survey Accuracy: Reducing Omissions in Survey Coverage

• 7 Point Cloud Utilization Methods to Improve the Accuracy of Deformation Surveys: Streamlining Reporting and Consensus Building

• Practical considerations for successfully leveraging point cloud data

• Summary

Why point clouds are gaining attention in deformation surveys

In condition surveys, the conventional workflow has been visual inspection, photography, recording with a scale, simple dimensional checks, and contact measurements when necessary. These methods are still effective, but the larger the area inspected and the more complex the structure’s geometry, the more likely omissions in records and subjective differences in assessment will occur. For example, one inspector may focus on crack width while another emphasizes surrounding deformation or out-of-plane displacement. Even when observing the same location, if there are differences in what is recorded and what is considered important, the resulting survey accuracy becomes inconsistent.

One reason point clouds are drawing attention is that they can potentially reduce this variability. Because they can record three-dimensional shapes as surfaces, areas that were overlooked at the time of the survey can be easily re-examined later by reviewing the recorded shapes. Photographs are viewpoint-dependent information, but point clouds carry positional and geometric information, making analyses such as slicing cross-sections to check deflection or viewing displacements from different directions possible. In other words, you can create a system that allows reassessment from recorded data instead of relying entirely on a single on-site observation.

Moreover, in condition surveys it is important not only to determine "whether abnormalities are present" but also "how much they have changed," "whether they have progressed since the previous inspection," and "in what form they appear including their surroundings." For that reason, there is an increasing need to grasp the condition as a surface rather than by single-point measurements. Point clouds are effective in this respect because they make surface information easy to handle and can capture localized distress and overall deformation simultaneously.

However, introducing point clouds does not automatically increase accuracy. If acquisition conditions are inappropriate, noise and missing data will increase, and you may actually make incorrect judgments. What matters is for what purpose, at what level of accuracy, and in which operations you will deploy and use the point clouds. Below, we will take a concrete look at uses that tend to be particularly effective in practice.

Method 1 for Leveraging Point Clouds to Improve Deformation Survey Accuracy: Standardize the Overall Shape for Recording

The first point is to standardize and record the overall shape of the object. In deformation surveys, if you focus only on anomalous areas, you may capture local symptoms but it becomes difficult to see how those deformations are positioned within the entire structure. For example, even if there is a crack in part of a wall, the assessment will differ depending on whether the whole wall is bulging forward or it is a localized shrinkage. By using point clouds, you can first capture the overall shape three-dimensionally and then locate the anomalous areas within that context.

This standardization is important because it provides a common reference surface that does not depend on the surveyor’s line of sight or how photographs are cropped. When taking photos on site, a crack photographed in close-up may appear clear, but its relationship to surrounding inclinations or offsets can become unclear. By contrast, with point clouds you can capture the condition of large surfaces — the entire wall surface, the entire floor surface, the entire slope, and the entire arch — and then confirm local anomalies. This reduces the risk of overestimating an isolated symptom or, conversely, overlooking it as part of an overall deformation.

In practice, simply photographing the object broadly is not enough. To capture the overall shape with the accuracy required for condition surveys, you need to design the acquisition coverage and density according to the survey target. For example, if you want to observe settlement or deflection, it is important to capture the surrounding intact areas as well so that relative comparisons can be made. If you want to see crack distribution trends, you must reduce blind spots so that surface continuity is clear. Even if you collect very high-density data only locally, the assessment will be unstable if the reference overall surface is coarse.

Furthermore, standardizing the overall shape makes it easier for multiple people to discuss the same data. Relying only on subjective on-site descriptions tends to result in vague communication such as "it's sticking out a bit" or "it's slightly tilted." However, if reference planes or reference lines are set on the point cloud and deformations are checked against them, the basis for discussion is aligned. The accuracy of a deformation survey includes not only the fineness of the measurements but also the reproducibility of judgments and their explainability. In that sense, standardizing the overall shape is the first practical approach to establish.

Point Cloud Utilization Method 2 for Improving the Accuracy of Deformation Surveys: Share Deformation Locations as Coordinates

The second point is to share the locations of anomalies using coordinates. A surprisingly common issue at sites conducting anomaly surveys is that it can be difficult to tell exactly where the previously noted spots were. Even if you mark them on photos, you may not be able to re-identify precisely the same location during the next inspection. This is especially true for long structures or areas where similar shapes repeat, where relying on memory or landmarks has its limits.

Point clouds can handle positional and shape information simultaneously, making it easier to spatially identify areas of deformation. This is not simply a matter of recording coordinate values. It is important to be able to share, within three-dimensional space, which surface, at what height, and how far from which edge an anomaly is located. This allows discussions about re-inspection, repair planning, or third-party verification to start from the same point even if the person in charge changes.

Positional reproducibility is especially important when managing the progression of deformations. For example, if you want to see how previously identified settlement locations or step locations have changed six months or a year later, the comparison itself won’t be valid unless you can return to the exact same position. Clearly managing deformation locations using point clouds improves the accuracy of time-series comparisons. Conversely, if positional correspondence remains ambiguous, there is a risk that the comparison targets will be misaligned even before considering the magnitude of the changes.

Furthermore, the value of sharing positional information increases in organizations where field and office tasks are more separated. When the person who collected data on site and the person who evaluates it in the office are different, photos alone cannot fully convey the spatial sense of the site. However, if anomalous locations are positioned on a point cloud, stakeholders in remote locations can more easily grasp the specific positions of the subjects. This reduces the extent to which differences in field experience translate directly into differences in report quality, and helps improve the overall accuracy of investigations across the organization.

Point Cloud Utilization Method 3 for Improving Deformation Survey Accuracy: Check Deformation Amounts in Cross-Sections

The third point is to check deformation amounts using cross-sections. In condition surveys, there are many subtle deformations that are difficult to judge from appearance alone. Bulging of wall surfaces, unevenness of floors, swelling of slopes, and localized displacements within a tunnel void are all prone to perspective and shooting-angle effects in photographs, making them hard to quantify. In such cases, extracting arbitrary cross-sections from point clouds and inspecting them makes it easier to capture changes as shapes rather than as visual impressions.

The advantage of cross-section inspection is that you can trace a deformation as a line. For example, even if part of a wall appears to be bulging, taking a cross-section makes it easier to identify which position is the peak and how far it projects out of the plane. Likewise, with pavement settlement, a cross-section reveals the continuity of elevation differences with the surroundings, making it easier to determine whether it is a local step or a widespread deformation. In evaluating deformations, not only the maximum value at a single point but also the distribution and shape of the deformation are important factors to consider.

What matters here is the perspective of where to take the cross-section. Setting the cross-section to include not only the location where anomalies are likely to occur but also the surrounding intact areas increases the accuracy of comparisons. If you extract only the abnormal portion, it can be difficult to determine whether it is truly deformation or an original shape variation. Verifying the cross-section against the object’s design geometry and known reference lines greatly improves the reliability of the inspection.

Cross-sections are also effective as reporting material. Deformations that are hard to convey with photographs alone become easier to understand by showing the cross-sectional shape. Visualization through cross-sections is especially helpful when explaining to clients or managers who have not seen the site why a particular deformation was judged to be important. The accuracy of deformation surveys is supported not only by the ability to detect issues but also by the ability to communicate them appropriately. Cross-sectional information obtained from point clouds is a practical means of supporting both.

Point Cloud Application Method 4 to Improve Deformation Survey Accuracy: Assessing Progression by Time-Series Comparison

The fourth point is to judge progressiveness by comparing observations over time. What makes condition surveys difficult is not only determining whether abnormalities exist, but also discerning whether they are problems currently in progress or conditions that have been stable since the past. Cracks and deformations are not necessarily judged dangerous merely because they exist. What is important are the changes over time—whether the deterioration is expanding, whether shape changes are continuing, and whether new symptoms are appearing in the surrounding area.

By leveraging point clouds, you can compare three-dimensional data acquired at different times and more easily grasp trends in changes. With conventional photo comparisons, differences in shooting position, focal length, and lighting conditions can make it difficult to determine whether any change has occurred. Because point clouds allow the shape itself to be compared, they are less susceptible to the effects of differing conditions. Of course, acquisition accuracy and the quality of alignment are important, but if operations become stable, the accuracy of progression assessment can be greatly improved.

Time-series comparisons are particularly effective for deformations that tend to manifest as geometric differences, such as settlement, deflection, bulging, wear, and the spread of defects. The width of a crack itself may sometimes require separate detailed inspection, but point clouds are extremely useful for examining the relationship with displacements of the surrounding surface and changes in shape. For example, if something that was previously a localized step is found in the next survey to have unevenness extending across the surrounding area, the priority of assessment will change.

However, to succeed in time-series comparisons, it is essential to consciously keep the acquisition conditions as consistent as possible each time. If the survey area, viewpoints, density, or the way references are set differ significantly, it becomes difficult to tell whether differences are true changes or merely variations in acquisition conditions. In practice, it is important to establish acquisition rules assuming future comparisons from the outset. Point clouds are useful as a one-time record, but they demonstrate their true value in continuous surveys. If you want to make the assessment of change progression more reliable, you should plan operations that include accumulating comparable data.

Point Cloud Techniques to Improve Deformation Survey Accuracy 5: Improve Interpretation Accuracy by Combining with Photographs

The fifth point is to improve the accuracy of interpretation by combining photographs with point clouds. Point clouds are excellent at capturing shape, but that alone does not allow you to identify all defects. Differences in surface color, staining, rust streaks, water-leak marks, material boundaries, repair traces, and the appearance of fine cracks are often easier to understand from photographic information. Conversely, there are situations where photographs alone make it difficult to grasp depth or the amount of deformation. Therefore, in practice it is important not to consider point clouds and photographs as opposing options, but to combine them in a complementary way.

For example, when determining whether discoloration on a wall is due to a water leak or merely surface dirt, photographic information is a great help. On the other hand, point clouds are effective for checking whether there is any slight bulging or unevenness around it. By overlaying the two, correlations that were difficult to see individually become visible. If visual abnormalities and geometric abnormalities can be confirmed simultaneously, estimating the cause of the anomaly and determining its priority also becomes easier.

Also, in investigation reports, combining photographs and point clouds increases persuasiveness. Photographs are intuitive and easy to understand, but they have the weakness of tending to be biased toward local information. Point clouds, on the other hand, make it easier to show the overall context and positional relationships, but they can take time to interpret for stakeholders who are not familiar with the site. Therefore, structuring the presentation so that photographs show the phenomenon while point clouds supplement the objectivity of position and shape makes the information easier to convey.

What practitioners should keep in mind is not to take photos incidentally while acquiring point clouds, but to record them so that the point clouds and photos can be easily associated later. If it is unclear which area was photographed from which direction, or which anomaly corresponds to which point-cloud position, the two types of information cannot be fully utilized even if collected. To improve the accuracy of condition surveys, creating a state in which each piece of information can be linked and used for interpretation is more important than increasing the number of data types.

Point Cloud Utilization Method 6 for Improving Deformation Survey Accuracy: Reducing Gaps in Survey Coverage

The sixth point is to reduce omissions in the survey area. In condition surveys, teams often have to inspect a wide area within a limited time, so attention inevitably tends to concentrate on priority spots. As a result, checks around obvious abnormalities, hard-to-access locations, and upward- or downward-looking directions can be insufficient, leading to the need for follow-up surveys later. The advantage of point clouds is that they make it easy to recheck on data the areas that seemed to have been seen on site but were actually missed.

Of course, point clouds also have blind spots and missing data. However, if you capture them across an area, it becomes at least easier to understand which parts have been captured and which have not. This is extremely important. In photo-based surveys, work can end without a clear awareness of the places that were not photographed, but with point clouds the missing regions themselves are more visible, making it easier to decide on re-acquisition or supplementation. Improving survey accuracy is not only about measuring anomalies with high precision, but also about properly defining the extent that needs to be checked.

Especially for structures where deterioration occurs at multiple, dispersed locations, gaps in survey coverage can greatly affect the overall assessment. Rather than stopping after observing a single crack or delamination, it is necessary to look more broadly to see whether the same type of deterioration continues in the surrounding area, whether similar symptoms appear at symmetric positions, or whether there is a concentration around drainage routes or joints. Point clouds support this overall view and make it easier to balance local observation with confirmation of the survey extent.

Furthermore, the ability to leave records in a form that can later explain areas where no abnormalities were confirmed on site has significant value. Condition surveys are not only about finding abnormal locations but also about indicating the areas where no major abnormalities were observed at the time. Using point clouds to reduce gaps in the survey coverage will be advantageous for future comparisons and for accountability.

Point Cloud Utilization Method 7 for Improving Deformation Survey Accuracy: Streamline Reporting and Consensus Building

The seventh point is to streamline reporting and consensus-building. In the field of condition surveys, how the results are communicated is as important as the results themselves. Even if the surveyor perceives a risk, if the client, managers, designers, and contractors do not reach the same understanding, decisions on repairs or additional investigations will be delayed. Conversely, if explanations can be based on objective shape information, discrepancies in understanding among stakeholders can be reduced.

What makes point clouds useful is that they can represent deterioration spatially. Photographs alone tend to capture only localized symptoms and can make it difficult to convey to someone seeing them for the first time why those areas are important. However, if point clouds are used to organize and show the location within the whole, relationships with the surroundings, cross-sectional shapes, and time-series differences, it becomes easier to share the rationale behind the inspection staff's judgments. In particular, when multiple deteriorations may be interrelated, materials that convey the overall picture are extremely effective.

Also, in consensus-building, the question of "what degree is acceptable" can become a point of contention. If you rely solely on subjective expressions and personal experience, discussions tend to end up at an impasse. Point clouds are not a panacea, but they at least objectify the starting point of the discussion. Being able to see where and what kind of shape differences exist and how far they extend makes it easier to align the criteria for judgment.

Furthermore, it facilitates future re-inspections and post-repair verifications. If the point cloud data used in this report are properly organized, it will be easier to refer to past conditions during subsequent checks. The fact that it can be leveraged not as a one-off report but as an asset for ongoing management is a major advantage of using point clouds. Improving the accuracy of damage surveys leads not only to higher detection accuracy on site but also to more accurate organizational judgments and decision-making.

Practical considerations for successfully utilizing point cloud data

We have introduced seven use cases so far, but to actually achieve results it is necessary to be aware of several caveats. The first important point is to determine the level of accuracy that is necessary and sufficient for the objective. Even in condition surveys, the required data density and acquisition methods change depending on whether the goal is to capture broad-area trends or to examine fine surface anomalies in detail. Acquiring data that is heavier than necessary increases processing load and can make field operations unsustainable, while data that is too coarse will be insufficient as a basis for decision-making. It is important to strike a balance between the objective and operational feasibility.

Next, do not leave how reference frames are established ambiguous. When performing time-series comparisons or sharing positions, you need to make the acquisition range, reference planes, and the way coordinates are handled as consistent as possible each time. If your approach changes with every acquisition, it may seem like you are looking at the same object, but in reality the assumptions underlying the comparison will have been undermined. The value of a point cloud lies more in its reproducibility when used continuously than in how it looks in a single capture.

Also, it is important not to separate on-site checks from in-office checks. Point clouds can be reviewed again and again in post-processing, but there is information that can only be understood on site. For example, the surrounding environment, the feel of materials, the movement of water, traffic conditions, and safety constraints may not be adequately conveyed by three-dimensional data alone. For that reason, it is desirable to have a workflow that links and records on-site inspections with point cloud evaluations. By tying on-site observations to the data, the accuracy of later interpretation is improved.

Furthermore, it is important to be mindful of presenting the information in a way that is easy to understand not only for those handling point cloud data but also for those receiving the reports. The ultimate purpose of a condition survey is not to create point clouds, but to lead to appropriate maintenance and management decisions. Even if the data itself is sophisticated, it will not fully realize its value if stakeholders cannot read it. Combining cross-sections, comparison diagrams, location maps, and corresponding photographs, and organizing them so that anyone can follow the key issues, determines practical success.

Summary

The value of using point clouds in deformation surveys goes beyond merely recording objects in three dimensions. It includes standardizing the overall shape to locate local anomalies, sharing deformation locations as coordinates to make re-checking easier, capturing deformation amounts in cross sections, assessing progression through time-series comparisons, combining with photographs to improve interpretation accuracy, reducing gaps in survey coverage, and objectifying reporting and consensus building. By putting these practices into action, the accuracy of deformation surveys is greatly improved.

On the other hand, point clouds are not a method that will deliver results simply by being introduced. Only by designing them to include acquisition conditions suited to the purpose, operations premised on continuous comparison, integration with on-site observations, and methods of organizing information that can be communicated to stakeholders do they become a system that is effective in practice. What is important in deterioration surveys is not only finding anomalies, but creating a state in which nothing is overlooked, comparisons can be made, and explanations can be given. Point clouds become a powerful foundation for that.

To further improve the accuracy of deformation surveys, it is essential not to rely solely on point clouds but also to create an environment that enables efficient on-site position verification and coordinate acquisition. For example, if tasks such as locating the survey target, correlating it with known control points, confirming the same point upon revisit, and sharing recorded positions become smoother, the overall reproducibility of point cloud operations will increase. If you want to streamline such on-site coordinate verification, combining measures such as LRTK — a smartphone-mounted high-precision GNSS positioning device — can further enhance the reliability of records and the practicality of deformation surveys. Thinking of point-cloud-based shape capture and on-site high-precision position verification together will make future deformation surveys more practical and easier to use.